Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A method for determining effectiveness of a stripe set, comprising: binarizing a raw image to obtain a binary image, wherein the raw image comprises a stripe pattern formed by a flashing light source and the stripe pattern comprises a plurality of stripes; obtaining a plurality of first connected domains in the binary image, wherein the plurality of first connected domains each correspond to pixels connected to each other and having a pixel value equal to a first threshold value; selecting a plurality of target connected domains from the plurality of first connected domains, wherein each of the plurality of target connected domains corresponds to one of the plurality of stripes in the stripe pattern; determining a set of the target connected domains corresponding to the stripe pattern as the stripe set, thereby to determine the stripe pattern; calculating a center (u 0 , v 0 ) of the stripe pattern based on the stripe set; and determining whether the stripe set is an effective stripe set based on the center (u 0 , v 0 ) of the stripe pattern, comprising: judging whether the center (u 0 , v 0 ) of the stripe pattern satisfies at least one of the following: { u 0 - R 2 < 0 u 0 + R 2 > kwide - 1 v 0 - R 2 < 0 v 0 + R 2 > kheight - 1 wherein μ 0 and v 0 are coordinate values of the center of the stripe pattern, R is the maximum width of the target connected domain of the stripe set, kwide is the width of the binary image, kheight is the height of the binary image; if at least one is satisfied, determining that the stripe set is an ineffective stripe set; and if none are satisfied, determining that the stripe set is an effective stripe set.
This invention relates to image processing for structured light systems and addresses the problem of determining the quality or effectiveness of a captured stripe pattern. The method involves processing a raw image containing a stripe pattern generated by a flashing light source. First, the raw image is converted into a binary image by applying a threshold. This binary image is then analyzed to identify connected regions of pixels that share a specific value. A subset of these connected regions, referred to as target connected domains, are selected. Each selected target connected domain is identified as corresponding to an individual stripe within the original stripe pattern. The collection of these target connected domains forms the stripe set, which represents the detected stripe pattern. Next, the geometric center of this detected stripe pattern is calculated based on the identified stripe set. Finally, the effectiveness of the stripe set is determined by evaluating the calculated center coordinates against predefined boundary conditions. Specifically, the center is considered ineffective if it falls outside a defined region within the image boundaries, considering the maximum width of the detected stripes. If the center lies within these boundaries, the stripe set is deemed effective.
2. The method as claimed in claim 1 , wherein before the obtaining the plurality of first connected domains in the binary image, the method further comprises: receiving the raw image, wherein the raw image is obtained by a rolling shutter image sensor.
A method for processing images captured by a rolling shutter image sensor addresses distortions caused by motion during image acquisition. The method involves receiving a raw image from a rolling shutter image sensor, which captures image data sequentially line by line, leading to distortions when the sensor or the scene moves. To correct these distortions, the method first obtains a plurality of first connected domains in the binary image, which represent regions of interest or features extracted from the raw image. Before this step, the method includes preprocessing the raw image to mitigate rolling shutter artifacts. This preprocessing may involve aligning or compensating for the sequential capture nature of the rolling shutter, ensuring that subsequent feature extraction and domain identification are accurate. The method then processes these connected domains to reconstruct or enhance the image, improving visual quality and reducing motion-induced distortions. The approach is particularly useful in applications where rolling shutter cameras are used, such as in consumer electronics, surveillance, or automotive imaging, where motion artifacts can degrade image clarity.
3. The method as claimed in claim 1 , wherein the selecting the plurality of target connected domains from the plurality of first connected domains comprises: determining a width of each first connected domain; comparing each width with a preset width; and determining eligible first connected domains from the plurality of first connected domains as the target connected domains, based on the comparing result.
This invention relates to a method for selecting target connected domains from a plurality of first connected domains in a network or data processing system. The method addresses the challenge of efficiently identifying suitable connected domains for further processing or analysis, particularly in scenarios where domain selection is based on specific width criteria. The method involves determining the width of each first connected domain, which may represent a measure of the domain's size, capacity, or other relevant metric. Each determined width is then compared against a preset width threshold. Based on this comparison, eligible first connected domains are identified as target connected domains. The preset width threshold serves as a filter to ensure only domains meeting specific criteria are selected, optimizing subsequent operations or analyses. The method may be applied in various contexts, such as network routing, data clustering, or domain partitioning, where selecting domains of appropriate size or capacity is critical. By dynamically comparing domain widths against a predefined threshold, the method ensures that only relevant domains are chosen, improving efficiency and accuracy in downstream processes. The approach is particularly useful in systems where domain selection impacts performance, resource allocation, or data integrity.
4. The method as claimed in claim 3 , wherein the comparing each width with a preset width and determining eligible first connected domains from the plurality of first connected domains comprises: taking x-axis in the direction of extension of a preset stripe image and y-axis perpendicular to the x-axis, based on an imaging surface of the rolling shutter image sensor; traversing all pixels of the plurality of first connected domains to obtain a maximum column coordinate y1 and a minimum column coordinate y2 of each of the pixels in the plurality of first connected domains; determining the width w of each first connected domain as follows: the maximum column coordinate y1 in the pixel minus the minimum column coordinate y2 in the pixel; judging whether the width w of each first connected domain meets the expression kh1≤w≤kh2, wherein kh1 is a preset first boundary condition parameter and kh2 is a preset second boundary condition parameter; and determining the first connected domain whose width meets the expression as the target connected domain.
This invention relates to image processing techniques for rolling shutter image sensors, specifically addressing the challenge of identifying target connected domains within an image based on width criteria. The method involves analyzing connected domains of pixels in a rolling shutter image to filter out those that meet predefined width constraints. The process begins by defining an x-axis along the direction of a preset stripe image and a y-axis perpendicular to it, aligned with the imaging surface of the sensor. For each connected domain, the method traverses all pixels to determine the maximum and minimum column coordinates (y1 and y2), calculating the width (w) as the difference between these coordinates. The width is then compared against preset boundary conditions (kh1 and kh2) to determine eligibility. Only domains where the width satisfies kh1 ≤ w ≤ kh2 are selected as target connected domains. This approach ensures accurate identification of relevant image regions based on their spatial dimensions, improving object detection and tracking in rolling shutter imaging systems. The technique is particularly useful in applications requiring precise analysis of image features under rolling shutter constraints.
5. The method as claimed in claim 4 , further comprising: deleting the first connected domain whose width fails to meet the expression, from the plurality of first connected domains.
A method for processing image data involves analyzing a plurality of first connected domains within an image to determine their widths. The method evaluates whether each connected domain meets a specified width criterion, which is defined by a mathematical expression. If a connected domain's width does not satisfy the expression, that domain is removed from the plurality of first connected domains. This process ensures that only connected domains meeting the width requirement are retained for further processing. The method may also involve generating a plurality of second connected domains from the remaining first connected domains, where each second connected domain is derived from a corresponding first connected domain. The second connected domains are then analyzed to determine their widths, and those that meet a second width criterion are retained. This approach is useful in image processing applications where specific structural features, such as lines or edges, must meet certain dimensional constraints to be considered valid. The method helps filter out irrelevant or noisy data, improving the accuracy of subsequent image analysis tasks.
6. The method as claimed in claim 2 , wherein binarizing the raw image comprises: filtering and binarizing the raw image to obtain a binary blurred image, which comprises a plurality of second connected domains, one of the second connected domains closest to the stripe pattern formed by a predetermined flashing light source is defined as a stripe pattern template; and wherein, after binarizing the raw image and before selecting the plurality of target connected domains, the method further comprises: identifying the target connected domains based on the stripe pattern template.
This invention relates to image processing techniques for analyzing raw images, particularly in applications involving structured light patterns, such as those used in 3D scanning or depth sensing. The problem addressed is accurately identifying and processing target regions within a raw image that may be obscured by noise or other artifacts, making it difficult to extract meaningful data from structured light patterns. The method involves binarizing a raw image to convert it into a binary blurred image, which contains multiple connected domains. The binarization process includes filtering the raw image to reduce noise and enhance the structured light pattern. One of the connected domains in the binary blurred image, specifically the one closest to the stripe pattern generated by a predetermined flashing light source, is designated as a stripe pattern template. This template serves as a reference for identifying target regions in subsequent processing steps. Before selecting the target connected domains, the method further includes an additional step of identifying these domains based on the stripe pattern template. This ensures that only relevant regions, which align with the expected structured light pattern, are considered for further analysis. The use of the template improves accuracy by filtering out irrelevant or noisy regions, thereby enhancing the reliability of the extracted data. This approach is particularly useful in applications where precise pattern recognition is critical, such as in 3D reconstruction or object detection.
7. The method as claimed in claim 2 , wherein after receiving the raw image and before binarizing the raw image, the method further comprises: filtering the raw image according to a filtering parameter to obtain a blurred image, wherein the filtering parameter is configured to merge the stripe pattern in the raw image; wherein binarizing the raw image to obtain the binary image comprises: binarizing the blurred image to obtain a binary blurred image; wherein, after binarizing the raw image and before selecting the plurality of target connected domains, the method further comprises: searching the binary blurred image to obtain a plurality of second connected domains; identifying shapes of the plurality of second connected domains; defining the second connected domain closest to the stripe pattern formed by a predetermined flashing light source, as a stripe pattern template; and identifying the plurality of target connected domains from the plurality of first connected domains, based on the stripe pattern template.
This invention relates to image processing techniques for improving the accuracy of detecting patterns in images, particularly in scenarios where stripe patterns from flashing light sources interfere with image analysis. The problem addressed is the difficulty in accurately identifying target objects in images when such stripe patterns obscure or distort the relevant features. The solution involves preprocessing the raw image to mitigate the interference caused by these patterns. The method begins by filtering the raw image using a configurable filtering parameter to blur the image, which merges the disruptive stripe pattern into the background. This filtered image is then binarized to produce a binary blurred image. Next, the binary blurred image is analyzed to identify connected domains, which are regions of contiguous pixels sharing similar properties. The shapes of these connected domains are examined to determine which one most closely resembles the expected stripe pattern from a known flashing light source. This identified domain is designated as a stripe pattern template. Using this template, the method then processes the original binary image to select target connected domains from the first set of connected domains detected in the raw image. The template helps distinguish relevant target objects from the interfering stripe patterns, ensuring more accurate detection and analysis. This approach enhances the reliability of image processing in applications where flashing light sources create unwanted artifacts.
8. The method as claimed in claim 7 , wherein the identifying the target connected domains from the plurality of first connected domains, based on the stripe pattern template comprises: judging whether the coordinates of pixels of the first connected domain are included in a set of coordinates of the pixels in the stripe pattern template; if so, determining the first connected domain as the target connected domain.
This invention relates to image processing, specifically identifying target regions within an image based on a predefined stripe pattern template. The problem addressed is accurately detecting specific connected domains in an image that match a given stripe pattern, which is useful in applications like document analysis, barcode recognition, or quality inspection. The method involves analyzing an image to extract a plurality of first connected domains, which are contiguous regions of pixels sharing similar properties. A stripe pattern template is used as a reference to identify target connected domains from these first connected domains. The process includes comparing the coordinates of pixels within each first connected domain against the coordinates of pixels in the stripe pattern template. If the coordinates of a first connected domain's pixels are included in the set of coordinates defined by the stripe pattern template, that first connected domain is classified as a target connected domain. This ensures that only regions matching the predefined stripe pattern are selected, improving accuracy in pattern recognition tasks. The method may be applied in automated systems requiring precise detection of structured patterns within images.
9. The method as claimed in claim 1 , wherein the raw image comprises a plurality of stripe patterns formed by a plurality of flashing light sources, the determining the set of the target connected domains corresponding to the stripe pattern as the stripe set comprises: classifying the plurality of target connected domains to obtain a plurality of stripe sets.
This invention relates to image processing techniques for analyzing raw images containing stripe patterns generated by flashing light sources. The problem addressed is accurately identifying and classifying connected regions (domains) in an image to determine which belong to specific stripe patterns, which is useful in applications like structured light 3D scanning or depth sensing. The method processes a raw image containing multiple stripe patterns formed by flashing light sources. The image includes various connected domains, which are regions of pixels grouped together based on similarity or continuity. The method classifies these domains to identify which ones correspond to the same stripe pattern, grouping them into distinct stripe sets. This involves analyzing the spatial and intensity characteristics of the domains to determine their association with specific stripes. The classification step ensures that domains belonging to the same stripe are grouped together, even if they are partially occluded or distorted. This improves the accuracy of subsequent processing steps, such as 3D reconstruction or depth mapping, by correctly associating image regions with their corresponding light sources. The technique is particularly useful in environments where multiple light sources are used to project structured patterns for sensing applications.
10. The method as claimed in claim 9 , wherein the classifying the plurality of target connected domains to obtain the plurality of stripe sets comprises: determining a plurality of initial stripe sets each including more than one of the target connected domains, wherein there are remaining target connected domains that fail to be classified into any of the plurality of initial stripe sets; selecting a jth target connected domain from the remaining target connected domains, wherein j is a positive integer; drawing a circle, with the center of mass of the jth target connected domain as the center and the length of the jth target connected domain as the diameter; judging whether the circle intersects any one of the target connected regions included in the plurality of initial stripe sets; and classifying the jth target connected region into the corresponding initial stripe set whose target connected regions intersect the circle, and updating the initial stripe sets to a stripe set.
This invention relates to a method for classifying target connected domains in a technical domain, such as image processing or pattern recognition, where the goal is to efficiently organize and group these domains into stripe sets. The problem addressed is the challenge of classifying all target connected domains into initial stripe sets while ensuring no domains remain unclassified. The method involves determining initial stripe sets, each containing multiple target connected domains, but some domains may remain unclassified. To resolve this, the method selects a remaining target connected domain, represented as the jth domain, and draws a circle centered at its center of mass with a diameter equal to the domain's length. The method then checks if this circle intersects any target connected regions within the initial stripe sets. If an intersection is found, the jth domain is classified into the corresponding initial stripe set, and the initial sets are updated to form a final stripe set. This iterative process ensures all domains are classified, improving the accuracy and completeness of domain grouping. The method is particularly useful in applications requiring precise domain classification, such as medical imaging, object detection, or data segmentation.
11. The method as claimed in claim 9 , wherein, after the classifying the plurality of target connected domains to obtain the plurality of stripe sets, the method further comprises: judging whether each of the plurality of stripe sets satisfies the following conditions: the number of the target connected domains in the stripe set is greater than a number threshold; and a length of the longest target connected domain of the stripe set is greater than a length threshold; deleting the stripe set which fails to satisfy both of the conditions, from the plurality of stripe sets.
This invention relates to data processing, specifically to methods for classifying and filtering connected domains in a data structure. The problem addressed is efficiently organizing and refining sets of connected domains to improve data processing performance or accuracy. The method involves classifying a plurality of target connected domains into multiple stripe sets based on predefined criteria. After classification, each stripe set is evaluated against two conditions: the number of target connected domains in the set must exceed a specified number threshold, and the length of the longest target connected domain in the set must exceed a specified length threshold. Any stripe set that fails to meet both conditions is removed from the plurality of stripe sets. This filtering step ensures that only relevant or significant stripe sets are retained, optimizing subsequent data processing tasks. The method may be applied in various domains, such as network analysis, data compression, or pattern recognition, where efficient domain classification and filtering are critical.
12. The method as claimed in claim 1 , wherein the calculating the center of the stripe pattern based on the stripe set comprises: calculating an average of the coordinate values of the centers of the target connected domains of the stripe set, to obtain the coordinate values of the center of the stripe pattern.
This invention relates to image processing techniques for analyzing stripe patterns, particularly in applications such as optical metrology or machine vision. The problem addressed is accurately determining the center of a stripe pattern, which is essential for precise measurements or alignment tasks. Stripe patterns often consist of multiple connected domains (regions of interest within the stripes), and conventional methods may struggle with noise or irregularities in these domains, leading to inaccurate center calculations. The method involves calculating the center of a stripe pattern by first identifying a set of target connected domains within the stripe pattern. Each connected domain represents a distinct region of the stripe. The method then computes the coordinate values of the centers of these individual connected domains. To determine the overall center of the stripe pattern, the method calculates the average of these coordinate values. This averaging process smooths out variations caused by noise or irregularities in the individual domains, resulting in a more accurate and robust center point for the entire stripe pattern. The approach is particularly useful in applications requiring high precision, such as semiconductor inspection, optical alignment, or automated quality control.
13. The method as claimed in claim 12 , wherein the calculating the average of the coordinate values of the centers of the target connected domains of the stripe set, to obtain the coordinate value of the center of the stripe pattern comprises: calculating an average of the coordinate values of the centers of the target connected domains of the stripe set according to the formula to obtain the center (u 0 , v 0 ) of the stripe pattern: ( u 0 , v 0 ) = ( ∑ i = 1 L i ( u min i + W i 2 ) L i , ∑ i = 1 L i ( v min i + H i 2 ) L i ) wherein i is the sequence number of the target connected domain of the stripe set, L i is the number of target connected domains of the stripe set, u min i is the minimum abscissa of the ith target connected domain, v min i is the minimum ordinate of the ith target connected domain, W i is the width of the ith target connected domain, and H i is the length of the ith target connected domain.
In the field of image processing, particularly in pattern recognition and analysis, a method is used to determine the center of a stripe pattern within an image. The problem addressed involves accurately identifying the central coordinates of a stripe pattern composed of multiple connected domains, which may vary in size and shape. The method calculates the average of the coordinate values of the centers of these target connected domains to derive the overall center of the stripe pattern. Specifically, the calculation involves summing the minimum abscissa (u_min_i) and half the width (W_i/2) of each target connected domain, then dividing by the total number of domains (L_i) to obtain the x-coordinate (u_0) of the center. Similarly, the y-coordinate (v_0) is derived by summing the minimum ordinate (v_min_i) and half the length (H_i/2) of each domain, then dividing by the total number of domains. This approach ensures precise localization of the stripe pattern's center, accounting for variations in individual domain dimensions. The method is particularly useful in applications requiring high-accuracy pattern detection, such as industrial inspection, medical imaging, or automated quality control.
14. The method as claimed in claim 13 , wherein after the calculating the average of the coordinate values of the centers of the target connected domains of the stripe set, and before the determining whether the stripe set is an effective stripe set based on the center of the stripe pattern, the method further comprises: filtering the pixel values of the pixels in a rectangular region Q to obtain the filtered pixel values, wherein the center of the rectangular region Q is the center (u 0 , v 0 ) of the stripe pattern, and the width of the rectangular region Q is the width of the maximum connected domain in the stripe set; and recalculating coordinate values of the center (u 0 , v 0 ) of the stripe pattern, with the pixel value of the filtered pixel as the weight.
In the field of image processing, particularly in stripe pattern analysis, a method is disclosed for improving the accuracy of detecting and analyzing stripe patterns in images. The problem addressed is the presence of noise or distortions in the image that can affect the precise determination of the center of a stripe pattern, leading to errors in subsequent analysis. The method involves calculating the average of the coordinate values of the centers of target connected domains within a stripe set. Before determining whether the stripe set is an effective stripe set based on the center of the stripe pattern, the method includes additional steps to enhance accuracy. First, the pixel values within a rectangular region Q are filtered to obtain filtered pixel values. The center of this rectangular region Q is aligned with the initially calculated center (u0, v0) of the stripe pattern, and the width of the region Q is set to the width of the maximum connected domain in the stripe set. Then, the coordinate values of the center (u0, v0) of the stripe pattern are recalculated, using the filtered pixel values as weights. This refinement step helps mitigate the impact of noise and improves the reliability of the stripe pattern analysis. The method is particularly useful in applications requiring high-precision stripe detection, such as optical measurement systems or quality control in manufacturing.
15. The method as claimed in claim 14 , wherein the filtering the pixel values of the pixels in the rectangular region Q to obtain the filtered pixel values comprises: filtering the pixel values of the pixels in the rectangular region Q by a filter template according to the formula, to obtain the filtered pixel value Gray u j ,v j : Gray u j , v j = ∑ x = - n - 1 2 n - 1 2 ∑ y = - n - 1 2 n - 1 2 { grav ( u j + x , v j + y ) > kvalue grav ( u j + x , v j + y ) grav ( u j + x , v j + y ) ≤ kvalue 0 n * n wherein n is the number of rows and columns of the filter template, grav (u j+x ,u j+y ) is the pixel value, and kvalue is the binarization threshold.
This invention relates to image processing, specifically a method for filtering pixel values in a rectangular region of an image to enhance or analyze features. The method addresses the challenge of effectively processing pixel data to extract meaningful information, such as binarization or noise reduction, in a computationally efficient manner. The technique involves applying a filter template to a rectangular region Q of an image, where the template has dimensions defined by n rows and n columns. For each pixel in the region, the method evaluates neighboring pixels within the template area using a binarization threshold (kvalue). If a neighboring pixel's value exceeds the threshold, its original value is retained; otherwise, it is set to zero. The filtered pixel value is then computed as the sum of these processed values across the template. This approach allows for selective filtering based on pixel intensity, enabling applications such as edge detection, feature extraction, or noise suppression. The method ensures that only significant pixel values contribute to the final output, improving the accuracy of subsequent image analysis tasks. The filter template size (n) and the threshold (kvalue) can be adjusted to optimize performance for different image types and processing requirements.
16. The method as claimed in claim 14 , wherein recalculating the coordinate values of the center (u 0 , v 0 ) of the stripe pattern with the pixel value of the filtered pixel as the weight comprises: recalculating coordinate values of the center (u 0 , v 0 ) of the stripe pattern according to the formula: ( u 0 , v 0 ) = ∑ Q Gray u j , v j ( u j , v j ) ∑ Q Gray u j , v j wherein (μ j , v j ) are coordinate values of the pixel in the rectangular region Q, and Gray u j ,v j is the pixel value after the pixel (μ j , v j ) being filtered.
This invention relates to image processing techniques for accurately determining the center coordinates of a stripe pattern in an image. The problem addressed is the need for precise localization of stripe patterns, which is critical in applications such as optical metrology, machine vision, and quality inspection. Existing methods may suffer from inaccuracies due to noise or variations in pixel intensity, leading to errors in center coordinate calculations. The method involves recalculating the center coordinates (u₀, v₀) of a stripe pattern within a rectangular region Q of an image. The recalculation uses pixel values as weights to improve accuracy. Specifically, the center coordinates are determined using a weighted average formula where each pixel's contribution is proportional to its filtered pixel value. The formula sums the products of each pixel's filtered gray value and its coordinates, then divides by the sum of the filtered gray values. This approach enhances robustness against noise and ensures precise localization of the stripe pattern's center. The method is particularly useful in applications requiring high-precision pattern detection, such as industrial inspection, medical imaging, and automated optical systems. By leveraging filtered pixel values as weights, the technique minimizes errors caused by uneven lighting or image artifacts, resulting in more reliable center coordinate calculations.
17. A terminal, comprising a rolling shutter image sensor, a memory, and a processor coupled to the rolling shutter image sensor and the memory, respectively; wherein the memory is configured to store images captured by the rolling shutter image sensor and a plurality of instructions; the processor is configured to execute the plurality of instructions; when the processor executing the plurality of instructions, a method for determining effectiveness of a stripe set is performed, wherein the method comprises: binarizing a raw image to obtain a binary image, wherein the raw image is captured by the rolling shutter image sensor and comprises a stripe pattern formed by a flashing light source, and the stripe pattern comprises a plurality of stripes; obtaining a plurality of first connected domains in the binary image, wherein the plurality of first connected domains each correspond to pixels connected to each other and having a pixel value equal to a first threshold value; selecting a plurality of target connected domains from the plurality of first connected domains, comprising: determining a width of each of the plurality of first connected domains; comparing each width with a preset width; and determining eligible first connected domains as the target connected domains, based on the comparing result, wherein each of the plurality of target connected domains corresponds to one stripe of the stripe pattern; determining a set of the target connected domains corresponding to the stripe pattern as the stripe set, thereby to determine the stripe pattern; calculating a center (u 0 , v 0 ) of the stripe pattern; and determining whether the stripe set is an effective stripe set based on the center (u 0 , v 0 ) of the stripe pattern, comprising: judging whether the center (u 0 , v 0 ) of the stripe pattern satisfies at least one of the following: { u 0 - R 2 < 0 u 0 + R 2 > kwide - 1 v 0 - R 2 < 0 v 0 + R 2 > kheight - 1 wherein μ 0 and v 0 are coordinate values of the center of the stripe pattern, R is the maximum width of the target connected domain of the stripe set, kwide is the width of the binary image, kheight is the height of the binary image; if at least one is satisfied, determining that the stripe set is an ineffective stripe set; if none are satisfied, determining that the stripe set is an effective stripe set.
This invention relates to a terminal device equipped with a rolling shutter image sensor for analyzing stripe patterns generated by a flashing light source. The device captures raw images containing these patterns, which are then processed to assess the effectiveness of the detected stripes. The system binarizes the raw image to produce a binary image, where connected domains of pixels meeting a threshold value are identified. These domains are filtered based on width to select target domains corresponding to individual stripes. The selected domains form a stripe set, which is used to determine the center coordinates (u0, v0) of the stripe pattern. The effectiveness of the stripe set is evaluated by checking if the center lies within a valid region of the image. If the center falls outside the boundaries defined by the image dimensions and the maximum stripe width (R), the stripe set is deemed ineffective. This process ensures accurate detection and validation of stripe patterns for applications such as structured light sensing or 3D imaging. The terminal includes a processor and memory to execute the image analysis, storing captured images and instructions for the evaluation method.
18. An image processing device, comprising a memory, and a processor coupled to a rolling shutter image sensor and the memory, respectively; wherein the memory is configured to store images obtained by the rolling shutter image sensor and a plurality of instructions; the processor is configured to execute the plurality of instructions; when the processor executing the plurality of instructions, a method for determining effectiveness of a stripe set is performed, wherein the method comprises: binarizing a stored image to obtain a binary image, wherein the stored image is captured by the rolling shutter image sensor and comprises a stripe pattern formed by a flashing light source, the stripe pattern comprises a plurality of stripes; obtaining a plurality of first connected domains in the binary image, wherein the plurality of first connected domains each correspond to pixels connected to each other and having a pixel value equal to a first threshold value; selecting a plurality of target connected domains from the plurality of first connected domains, wherein each of the plurality of target connected domains corresponds to one of the plurality of stripes in the stripe pattern; determining a set of the target connected domains corresponding to the stripe pattern as the stripe set, to determine the stripe pattern; calculating a center (u 0 , v 0 ) of the stripe pattern; and determining whether the stripe set is an effective stripe set based on the center (u 0 , v 0 ) of the stripe pattern, comprising: judging whether the center (u 0 , v 0 ) of the stripe pattern satisfies at least one of the following: { u 0 - R 2 < 0 u 0 + R 2 > kwide - 1 v 0 - R 2 < 0 v 0 + R 2 > kheight - 1 wherein μ 0 to and v 0 are coordinate values of the center of the stripe pattern, R is the maximum width of the target connected domain of the stripe set, kwide is the width of the binary image, kheight is the height of the binary image; if at least one is satisfied, determining that the stripe set is an ineffective stripe set; if none are satisfied, determining that the stripe set is an effective stripe set.
This invention relates to image processing for evaluating the effectiveness of stripe patterns captured by a rolling shutter image sensor, particularly in applications involving flashing light sources. The problem addressed is determining whether a detected stripe pattern is valid or corrupted, which is critical for accurate image analysis in systems like LiDAR or structured light imaging. The device includes a memory and a processor connected to a rolling shutter image sensor. The sensor captures images containing stripe patterns generated by a flashing light source. The processor performs a method to assess the stripe pattern's validity. First, the captured image is binarized to produce a binary image, where connected pixels above a threshold form connected domains. These domains are filtered to select those corresponding to individual stripes in the pattern. The selected domains form a stripe set, and the center of the stripe pattern is calculated. The stripe set's effectiveness is determined by checking if the center's coordinates (u0, v0) fall within a valid region. If the center is too close to the image edges (within a margin defined by the stripe width R and image dimensions kwide, kheight), the stripe set is deemed ineffective. Otherwise, it is considered valid. This ensures only reliable stripe patterns are used for further processing.
20. The terminal as claimed in claim 17 , wherein, comparing each width with a preset width and determining eligible first connected domains as the target connected domains, comprises: taking x-axis in the direction of extension of a preset stripe image and y-axis perpendicular to the x-axis, based on an imaging surface of the rolling shutter image sensor; traversing all pixels of the plurality of first connected domains to obtain a maximum column coordinate y1 and a minimum column coordinate y2 of each of the pixels in the plurality of first connected domains; determining the width w of each first connected domain as follows: the maximum column coordinate y1 in the pixel minus the minimum column coordinate y2 in the pixel; judging whether the width w of each first connected domain meets the expression kh1≤w≤kh2, wherein kh1 is a preset first boundary condition parameter and kh2 is a preset second boundary condition parameter; and determining the first connected domain whose width meets the expression as the target connected domain.
This invention relates to image processing in rolling shutter image sensors, specifically for identifying target connected domains within an image. The problem addressed is accurately selecting connected pixel regions that meet predefined width criteria, which is critical for applications like object detection or image segmentation. The method involves analyzing connected domains of pixels in a rolling shutter image. The image is processed with an x-axis aligned with the direction of a preset stripe image and a y-axis perpendicular to it. For each connected domain, the system traverses all pixels to determine the maximum and minimum column coordinates (y1 and y2). The width (w) of the domain is calculated as the difference between these coordinates (y1 - y2). The width is then compared against preset boundary conditions (kh1 and kh2). If the width satisfies kh1 ≤ w ≤ kh2, the domain is classified as a target connected domain. This ensures only domains within the specified width range are selected, improving accuracy in subsequent image analysis tasks. The approach leverages spatial pixel relationships to filter relevant regions efficiently.
Unknown
October 6, 2020
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.